Array programming with NumPy.

Array programming provides a powerful, compact and expressive syntax for accessing, manipulating and operating on data in vectors, matrices and higher-dimensional arrays. NumPy is the primary array programming library for the Python language. It has an essential role in research analysis pipelines in fields as diverse as physics, chemistry, astronomy, geoscience, biology, psychology, materials science, engineering, finance and economics. For example, in astronomy, NumPy was an important part of the software stack used in the discovery of gravitational waves1 and in the first imaging of a black hole2. Here we review how a few fundamental array concepts lead to a simple and powerful programming paradigm for organizing, exploring and analysing scientific data. NumPy is the foundation upon which the scientific Python ecosystem is constructed. It is so pervasive that several projects, targeting audiences with specialized needs, have developed their own NumPy-like interfaces and array objects. Owing to its central position in the ecosystem, NumPy increasingly acts as an interoperability layer between such array computation libraries and, together with its application programming interface (API), provides a flexible framework to support the next decade of scientific and industrial analysis.

Author Correction: SciPy 1.0: fundamental algorithms for scientific computing in Python.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

SciPy 1.0: fundamental algorithms for scientific computing in Python.

SciPy is an open-source scientific computing library for the Python programming language. Since its initial release in 2001, SciPy has become a de facto standard for leveraging scientific algorithms in Python, with over 600 unique code contributors, thousands of dependent packages, over 100,000 dependent repositories and millions of downloads per year. In this work, we provide an overview of the capabilities and development practices of SciPy 1.0 and highlight some recent technical developments.

Teaching Computational Reproducibility for Neuroimaging.

We describe a project-based introduction to reproducible and collaborative neuroimaging analysis. Traditional teaching on neuroimaging usually consists of a series of lectures that emphasize the big picture rather than the foundations on which the techniques are based. The lectures are often paired with practical workshops in which students run imaging analyses using the graphical interface of specific neuroimaging software packages. Our experience suggests that this combination leaves the student with a superficial understanding of the underlying ideas, and an informal, inefficient, and inaccurate approach to analysis. To address these problems, we based our course around a substantial open-ended group project. This allowed us to teach: (a) computational tools to ensure computationally reproducible work, such as the Unix command line, structured code, version control, automated testing, and code review and (b) a clear understanding of the statistical techniques used for a basic analysis of a single run in an MR scanner. The emphasis we put on the group project showed the importance of standard computational tools for accuracy, efficiency, and collaboration. The projects were broadly successful in engaging students in working reproducibly on real scientific questions. We propose that a course on this model should be the foundation for future programs in neuroimaging. We believe it will also serve as a model for teaching efficient and reproducible research in other fields of computational science.

Learning from open source software projects to improve scientific review.

Peer-reviewed publications are the primary mechanism for sharing scientific results. The current peer-review process is, however, fraught with many problems that undermine the pace, validity, and credibility of science. We highlight five salient problems: (1) reviewers are expected to have comprehensive expertise; (2) reviewers do not have sufficient access to methods and materials to evaluate a study; (3) reviewers are neither identified nor acknowledged; (4) there is no measure of the quality of a review; and (5) reviews take a lot of time, and once submitted cannot evolve. We propose that these problems can be resolved by making the following changes to the review process. Distributing reviews to many reviewers would allow each reviewer to focus on portions of the article that reflect the reviewer's specialty or area of interest and place less of a burden on any one reviewer. Providing reviewers materials and methods to perform comprehensive evaluation would facilitate transparency, greater scrutiny, and replication of results. Acknowledging reviewers makes it possible to quantitatively assess reviewer contributions, which could be used to establish the impact of the reviewer in the scientific community. Quantifying review quality could help establish the importance of individual reviews and reviewers as well as the submitted article. Finally, we recommend expediting post-publication reviews and allowing for the dialog to continue and flourish in a dynamic and interactive manner. We argue that these solutions can be implemented by adapting existing features from open-source software management and social networking technologies. We propose a model of an open, interactive review system that quantifies the significance of articles, the quality of reviews, and the reputation of reviewers.

Data sharing for computational neuroscience.

Computational neuroscience is a subfield of neuroscience that develops models to integrate complex experimental data in order to understand brain function. To constrain and test computational models, researchers need access to a wide variety of experimental data. Much of those data are not readily accessible because neuroscientists fall into separate communities that study the brain at different levels and have not been motivated to provide data to researchers outside their community. To foster sharing of neuroscience data, a workshop was held in 2007, bringing together experimental and theoretical neuroscientists, computer scientists, legal experts and governmental observers. Computational neuroscience was recommended as an ideal field for focusing data sharing, and specific methods, strategies and policies were suggested for achieving it. A new funding area in the NSF/NIH Collaborative Research in Computational Neuroscience (CRCNS) program has been established to support data sharing, guided in part by the workshop recommendations. The new funding area is dedicated to the dissemination of high quality data sets with maximum scientific value for computational neuroscience. The first round of the CRCNS data sharing program supports the preparation of data sets which will be publicly available in 2008. These include electrophysiology and behavioral (eye movement) data described towards the end of this article.